[R-sig-ME] time-series analysis: how to deal with 'nuisance'factors influencing a trend?

David Duffy davidD at qimr.edu.au
Mon Dec 20 04:55:23 CET 2010

On Wed, 15 Dec 2010, Giancarlo Sadoti wrote:

> My central question relates to the apparent and confounding influence of 
> another characteristic of each site (in this case the mean # of 
> individuals) on the "trend" fixed effect (in this case the change in 
> per-site # of individuals across the three years).  How is the best way 
> to 'control' for this influence in a mixed model in order to get to the 
> 'true' trend?

Your lmer(COUNT~YEAR+YEAR:MEAN_COUNT+(1|SITE),family=poisson, data=data)
is probably how I would do it, given you say diagnostics suggest a 
Poisson model for the counts was OK.  I would look to biology re 
alternative models: the generation time for your 
species is longer than YEAR? you expect larger populations to be more 

Cheers, David.

| David Duffy (MBBS PhD)                                         ,-_|\
| email: davidD at qimr.edu.au  ph: INT+61+7+3362-0217 fax: -0101  /     *
| Epidemiology Unit, Queensland Institute of Medical Research   \_,-._/
| 300 Herston Rd, Brisbane, Queensland 4029, Australia  GPG 4D0B994A v

More information about the R-sig-mixed-models mailing list